This chapter covers Azure VM Scale Sets, a service for managing multiple virtual machines with automatic scaling. AZ-900 tests this under objective 2.2, as a compute service that demonstrates scaling and load distribution concepts in practice.
Jump to a section
A simple way to picture Azure VM Scale Sets
A single chef can handle a quiet evening, but during a dinner rush, you need more chefs — and after the rush, you scale back to save on labor costs. Azure VM Scale Sets work similarly: they automatically create and destroy virtual machine instances based on demand, so you have just enough capacity without wasting money on idle machines when demand is low.
What VM Scale Sets do
Azure VM Scale Sets manage a group of identical virtual machines, automatically increasing or decreasing the number of instances based on demand, without manual intervention. This combines scalability (handling varying demand) and availability (distributing load across multiple instances).
How autoscaling works
Rules define when to add or remove instances based on metrics like CPU usage or request count. When the metric exceeds a threshold (for example, "if average CPU > 75%"), the scale set adds instances; when it drops below another threshold, instances are removed, reducing cost during low-demand periods.
Load balancing
When multiple instances exist in a scale set, incoming traffic is distributed across them, so no single instance becomes overwhelmed.
How it compares to a single VM
A single VM has fixed capacity — if demand exceeds what it can handle, performance suffers and scaling requires manual intervention (usually deleting and recreating the VM with a larger size). A scale set handles varying demand automatically and distributes load across multiple instances.
Define the VM configuration
The scale set is configured with the desired VM size, operating system image, and other settings — all instances created by the scale set will use this configuration.
Set scaling rules
Rules are defined specifying when to scale up (add instances) or scale down (remove instances) based on specific metrics.
Configure load balancing
Load balancing is configured so incoming traffic is distributed across instances rather than going to just one.
Monitor and adjust
Scaling happens automatically based on the defined rules, but monitoring confirms the rules are working as intended.
A website uses a VM scale set to automatically add instances during peak traffic hours and remove them during quiet periods, maintaining performance while minimizing cost by only paying for capacity actually in use.
Objective 2.2 expects candidates to understand VM Scale Sets as a managed service for handling variable demand through automatic scaling and load distribution.
A common wrong answer is treating VM Scale Sets as simply "a set of VMs" without understanding the automatic scaling aspect — the key differentiator is the automation and dynamic responsiveness to demand.
Stable terms: VM Scale Set, autoscaling, load balancing. Memory trick: VM Scale Sets automate scaling and load distribution, so you don't need to manually add/remove VMs or route traffic yourself.
Azure VM Scale Sets manage a group of identical VMs with automatic scaling based on demand.
Instances are automatically added during high demand and removed during low demand, reducing costs.
Load balancing distributes traffic across instances, preventing any single instance from becoming overwhelmed.
Scaling rules are defined based on metrics like CPU usage, automating the response to demand changes.
These come up on the exam all the time. Here's how to tell them apart.
Single Virtual Machine
Fixed capacity
Manual scaling required
No automatic load distribution
VM Scale Set
Automatic scaling based on demand
Reduces cost during low-demand periods
Automatic load distribution across instances
Mistake
A VM Scale Set is just a fixed collection of VMs.
Correct
The defining trait is automatic scaling — the scale set automatically adds or removes instances based on demand, not a fixed number.
Mistake
VM Scale Sets only work with Windows VMs.
Correct
They work with both Windows and Linux virtual machines.
It's a service that manages a group of identical virtual machines, automatically adding or removing instances based on demand, without manual intervention.
Rules based on metrics like CPU usage define when to add instances (when usage exceeds a threshold) or remove them (when usage drops below another threshold), automatically responding to demand without manual intervention.
Incoming traffic is automatically distributed across instances in the scale set, so no single instance becomes a bottleneck — if one instance fails, the others continue handling traffic.
You've just covered Azure VM Scale Sets — now see how well it sticks with free AZ-900 practice questions. Full explanations included, no account needed.
Done with this chapter?